From a69052be7f9a4008e2b09578e9db5fdebc186111 Mon Sep 17 00:00:00 2001 From: Brian Behlendorf Date: Fri, 1 Oct 2010 16:54:52 -0700 Subject: Initial zio delay timing While there is no right maximum timeout for a disk IO we can start laying the ground work to measure how long they do take in practice. This change simply measures the IO time and if it exceeds 30s an event is posted for 'zpool events'. This value was carefully selected because for sd devices it implies that at least one timeout (SD_TIMEOUT) has occured. Unfortunately, even with FAILFAST set we may retry and request and not get an error. This behavior is strongly dependant on the device driver and how it is hooked in to the scsi error handling stack. However by setting the limit at 30s we can log the event even if no error was returned. Slightly longer term we can start recording these delays perhaps as a simple power-of-two histrogram. This histogram can then be reported as part of the 'zpool status' command when given an command line option. None of this code changes the internal behavior of ZFS. Currently it is simply for reporting excessively long delays. --- include/sys/fm/fs/zfs.h | 2 ++ include/sys/zio.h | 6 ++++++ 2 files changed, 8 insertions(+) (limited to 'include') diff --git a/include/sys/fm/fs/zfs.h b/include/sys/fm/fs/zfs.h index d157ea153..4571939d2 100644 --- a/include/sys/fm/fs/zfs.h +++ b/include/sys/fm/fs/zfs.h @@ -35,6 +35,7 @@ extern "C" { #define FM_EREPORT_ZFS_CHECKSUM "checksum" #define FM_EREPORT_ZFS_IO "io" #define FM_EREPORT_ZFS_DATA "data" +#define FM_EREPORT_ZFS_DELAY "delay" #define FM_EREPORT_ZFS_CONFIG_SYNC "config.sync" #define FM_EREPORT_ZFS_POOL "zpool" #define FM_EREPORT_ZFS_POOL_DESTROY "zpool.destroy" @@ -81,6 +82,7 @@ extern "C" { #define FM_EREPORT_PAYLOAD_ZFS_ZIO_OFFSET "zio_offset" #define FM_EREPORT_PAYLOAD_ZFS_ZIO_SIZE "zio_size" #define FM_EREPORT_PAYLOAD_ZFS_ZIO_FLAGS "zio_flags" +#define FM_EREPORT_PAYLOAD_ZFS_ZIO_DELAY "zio_delay" #define FM_EREPORT_PAYLOAD_ZFS_PREV_STATE "prev_state" #define FM_EREPORT_PAYLOAD_ZFS_CKSUM_EXPECTED "cksum_expected" #define FM_EREPORT_PAYLOAD_ZFS_CKSUM_ACTUAL "cksum_actual" diff --git a/include/sys/zio.h b/include/sys/zio.h index 97d8ec74d..a46918174 100644 --- a/include/sys/zio.h +++ b/include/sys/zio.h @@ -116,6 +116,11 @@ enum zio_compress { ZIO_COMPRESS_ON_VALUE == ZIO_COMPRESS_LZJB) || \ (compress) == ZIO_COMPRESS_OFF) +/* + * Default Linux timeout for a sd device. + */ +#define ZIO_DELAY_MAX (30 * MILLISEC) + #define ZIO_FAILURE_MODE_WAIT 0 #define ZIO_FAILURE_MODE_CONTINUE 1 #define ZIO_FAILURE_MODE_PANIC 2 @@ -401,6 +406,7 @@ struct zio { enum zio_flag io_orig_flags; enum zio_stage io_orig_stage; enum zio_stage io_orig_pipeline; + uint64_t io_delay; int io_error; int io_child_error[ZIO_CHILD_TYPES]; uint64_t io_children[ZIO_CHILD_TYPES][ZIO_WAIT_TYPES]; -- cgit v1.2.3